Investigative Ophthalmology & Visual Science, Vol. 33, No. 2, February 1992
Copyright © Association for Research in Vision and Ophthalmology
Total Retinal Volumetric Blood Flow Rate in Diabetic
Patients With Poor Glycemic Control
Juan E. Grunwald, Charles E. Riva, Joan Baine, and Alexander J. Brucker
Total retinal volumetric blood flow rate was measured in 12 normal subjects and 18 poorly controlled
diabetic patients with background diabetic retinopathy. Maximum or center-line erythrocyte velocity
(Vmax) was assessed by bidirectional laser Doppler velocimetry in four to five major retinal veins of one
eye of each subject. Venous diameter (D) was measured from monochromatic fundus photographs.
Total venous cross-section and measured total retinal volumetric blood flow in the diabetic patients
were significantly larger than normal (P = 0.001 and P = 0.02, respectively). A positive linear correlation was found between Vmax and D in normal and diabetic eyes. Volumetric blood flow rate, Q, varied
with D at a power of 2.87 in normal eyes and 2.67 in diabetic eyes. Total volumetric blood flow
correlated with total venous cross-section. It was found that Q in the temporal retina was significantly
larger than in the nasal retina in normal subjects {P = 0.0008) and diabetic patients (P = 0.0002). A
significant difference in Q was observed between the superior and inferior retina in diabetic patients (P
= 0.03) but not in normal subjects. The retinal vascular regulatory response to 100% oxygen breathing
was reduced (P = 0.019) in diabetic patients and correlated with the level of background diabetic
retinopathy. A close estimate of total volumetric blood flow may be obtained from blood flow measurement in one major retinal vein and the determination of total venous cross-section. This may be
important for clinical studies in which measurements of all individual retinal veins may not be feasible.
Invest Ophthalmol Vis Sci 33:356-363,1992
The retina is one of the major sites of the body
affected by diabetic microvascular pathology. Abnormalities of the retinal circulation and its regulatory
responses1"12 are known to occur in the eyes of diabetic patients with retinopathy and are probably an
important factor in the development of retinal diabetic pathology.
Laser Doppler velocimetry studies of retinal hemodynamics in diabetes mellitus have been done previously.6"11 In most of these studies, measurements of
one single retinal vessel have been obtained in each
eye. Because volumetric blood flow rate is a function
of the size of the vessel measured and the area perfused, large variability in blood flow between patients
has been reported.8"11 In our study, to overcome this
problem, we determined total retinal venous volumetric blood flow rate by adding the flows measured in
four to five major retinal veins in each eye.
Increased blood glucose levels and large fluctuations in these levels are associated with an increased
incidence of diabetic retinopathy13'14 and are known
to affect retinal hemodynamics in diabetic patients.9
To characterize these retinal hemodynamic abnormalities further, we determined total retinal venous volumetric blood flow and the vascular regulatory response to hyperoxia in a group of poorly controlled
patients with diabetes mellitus and compared the results with those obtained in normal control subjects.
Materials and Methods
Eighteen patients with insulin-dependent diabetes
mellitus (age range, 18-38 yr; mean ± standard deviation, 29 ± 5 yr) were included in this study. The duration of diabetes ranged from 6-28 yr (mean, 17 ± 7
yr). All patients had background diabetic retinopathy,
an otherwise normal eye examination, and glycosylated hemoglobin (GHb) values higher than three
standard deviations above the mean for nondiabetic
subjects. Average GHb measured by affinity chromatography was 12.4 ± 3.3% (upper limit of normal
range, 8%). Blood glucose was determined from finger
capillary blood samples using an Accu-Check blood
glucose monitor (Boehringer Mannheim, Indianapolis, IN). The average blood glucose at the time of retinal volumetric blood flow determination was 192
± 100 mg/dl (Table 1).
From the Department of Ophthalmology, Scheie Eye Institute,
School of Medicine, University of Pennsylvania, Philadelphia,
Pennsylvania.
Supported by National Institutes of Health grants EY05775,
EY03242, and RR-00040 (Bethesda, Maryland) and the Vivian
Simkins Lasko Retinal Vascular Research Fund.
Submitted for publication: November 20, 1990; accepted September 17, 1991.
Reprint requests: Dr. Juan E. Grunwald, Scheie Eye Institute, 51
North 39th Street, Philadelphia, PA 19104.
356
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
No. 2
TOTAL RETINAL DLOOD FLOW IN DIABETES / Grunwold er ol
357
Table 1. Subject characteristics
Normal
Age (yr)
Disease duration (yr)
Glycosylated hemoglobin (%)
Blood glucose (mg/dl)
Mean brachial blood pressure (mmHg)
Intraocular pressure (mmHg)
Mean ± SD
Range
Mean ± SD
Range
Significance*
32 ± 7
19-45
53-109
71-100
12-18
18-39
6-28
7.3-18.7
63-400
72-108
8-18
NS
78 ± 17
85 ± 10
15±2
29 ± 6
17±7
12.4 ±3.3
192 ± 100
84 ± 11
15±3
NS = not statistically significant.
Excluded from the study were patients who had: (1)
previous treatment with three or more daily injections
of insulin or an insulin pump; (2) three or more documented episodes of diabetic ketoacidosis requiring
hospitalization; (3) history of systemic hypertension
or substance abuse; (4) obesity, denned as body
weight larger than 130% of ideal body weight; (5) presence of intraocular disease other than diabetic retinopathy; or (6) previous laser photocoagulation treatment. Results in diabetic patients were compared
with those obtained from 12 normal volunteers (age
range, 19-45 yr; mean, 32 ± 7) whose blood glucose
level at the time of retinal volumetric flow measurements was an average of 78 ± 17 mg/dl (Table 1).
All eyes studied had a best-refracted visual acuity of
6/7.5 or better, an intraocular pressure < 21 mm Hg,
and a normal slit-lamp examination. A description
and statistical comparison of the characteristics of
normal subjects and diabetic patients is provided in
Table 1. After a detailed explanation of the procedures, all subjects were asked to sign an appropriate
consent form approved by the Internal Review Board
of our institution.
Only one eye, chosen at random, was investigated
in each subject. After pupil dilation with tropicamide
1% and phenylephrine hydrochloride 10%, a Polaroid
(Cambridge, MA) color fundus photograph of the posterior fundus was obtained to localize the sites of bidirectional laser Doppler velocimetry (BLDV) measurements. These measurements of the maximum,
center-line erythrocyte velocity (Vmax) were obtained
in four to five major retinal veins. Velocity was measured on straight portions of veins at a distance of less
than 2 disc diameters from the center of the optic
nerve head. We avoided sites close to venous junctions or arteriovenous crossings and those where two
vessels lay close to each other. The location of the
measurement site was marked on the Polaroid photograph for later reference. We usedflowmeasurements
from veins instead of arteries because the minimal
flow pulsatility in these vessels permitted a more accurate determination of the average velocity.815
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
Diabetes
• Nonpaired
P = 0.001
NS
NS
student t-test.
During the BLDV measurements, an area of the
posterior retina (30° in diameter) was illuminated at a
wavelength of 570 nm with a retinal irradiance of
about 0.30 mW/cm2. The levels of laser light used
during the experiments were within the maximum
permissible levels for extended sources.16
Fundus photographs were taken in monochromatic
light at 570 nm using a Zeiss (Oberkochen, Germany)
fundus camera and Plus-X pan film (Eastman Kodak,
Rochester, NY). Intraocular pressure was measured
by applanation tonometry, and brachial artery blood
pressure was obtained by sphygmomanometry.
Volumetric blood flow rate (Q) was calculated as
described previously89 as Vmean• TTD2/4, where mean
blood velocity (Vmean) was calculated as C • Vmax. A
value for C equal to 1/1.6 was used17, and the relationship between Vmax and Vmean was assumed to be the
same in normal and diabetic subjects. This assumption has been discussed previously.8 The venous diameter at the site of BLDV measurement, D (determined
from projected photographic negatives), was an average obtained from six photographs. Total venous
cross-section (ST) was calculated by adding the crosssection of all visible veins observed around the disc.
Measurements of Vmax and D were done on a major
retinal vein during room air breathing and during 4-6
min of breathing 100% oxygen at atmospheric pressure. The retinal vascular regulatory response to hyperoxia (R), defined as the percentage decrease in Q
between air and 100% oxygen breathing, was calculated using the formula: R = 100 (Qair - Qox)/Qair.
All measurements of D were done by one examiner, and all Vmax determinations were done by another one. Both examiners were masked with regard
to the results obtained by the other and the clinical
status of the subject measured.
In the diabetic subjects, seven standard field stereo
color fundus photographs were obtained as in the
Early Treatment of Diabetic Retinopathy Study
(ETDRS) protocol. Retinopathy was assessed in a
masked fashion at the Fundus Photographic Reading
Center of the University of Wisconsin. An overall reti-
358
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1992
nopathy level according to the ETDRS grading protocol was assigned to the study eye of each patient.
Mean brachial artery blood pressure (BPm) was calculated as BPd + V3(BPS - BPd), where BPS and BPd are
the brachial artery systolic and diastolic pressures.
Paired and unpaired two-tailed student t-tests,
correlation analysis based on r2 values obtained from
regression fits of the data, and rank correlation analysis were used in the evaluation of the results. All data
in which student t-tests were applied were validated
for normality using the Wilk-Shapiro test. A Wilcoxon rank-sum test was used in those cases in which
the data were not distributed normally. We used P
< 0.05 for statistical significance.
Results
Table 1 shows the studied subjects' characteristics.
There were no significant differences in age, BPm, or
intraocular pressure between diabetic patients and
normal subjects. Blood glucose determined at the
time of BLDV was significantly higher than normal in
diabetic patients (P = 0.001, by unpaired student ttest).
Figures 1 and 2 illustrate the dependence of Vmax
and Q on the D of each of the four to five individual
veins measured in normal subjects and diabetic patients. The D ranged from 72-185 ixm in normal subjects and from 87-226 nm in diabetic patients. When
all points shown in Figure 1 were considered as independent data points, which they were not, a positive
correlation was found between Vmax and D in normal
subjects (correlation coefficient, r = 0.75) and in dia1
1
Normals
Diabetics
0.0
100
120
140
160
180
Vessel Diameter (/xm)
Fig. 1. Relationship between maximum erythrocyte velocity
(Vmax) and venous diameter (D), based on measurements obtained
from four to five largest retinal veins in eyes of normal subjects
(squares) and diabetic patients (circles). Correlation coefficients, r,
for the linear fit are 0.75, (linear regression equation Vmax = 0.14
+ 0.01 X D) for normals, and 0.37 (linear regression equation Vmax
= 0.45 + 0.01 X D) for diabetic patients when all points are considered as independent observations.
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
Vol. 33
'Normals
•Diabetics
0
20
40
60
80
100 120
140
160
180 200 220 240
Vessel Diameter (/Am)
Fig. 2. Relationship between volumetric blood flow (Q) and venous diameter (D), based on measurements obtained from four to
five largest retinal veins in eyes of normal subjects (squares) and
diabetic patients (circles). Correlation coefficients for the power
curve fit are r = 0.96 for normals, and r = 0.83 for diabetic patients
when all points are considered simultaneously as independent observations. Q varies as 6.1 X 10"6 D 287 in normals and as 1.4 X 10"5
D267 in diabetic patients.
betic patients (r = 0.37), showing that erythrocyte velocity increases with increasing D. When the relationship between Vmax and D was investigated in each
subject separately, a positive correlation with an average r = 0.88 ± 0.12 (range, 0.57-1.00) was seen in
normal subjects and an average r = 0.71 ± 0.27
(range, 0.1-0.99) in diabetic patients. The average
correlation coefficient was not significantly lower
than normal in diabetic patients (P > 0.05, by Wilcoxon rank-sum test).
The correlation coefficient between Q and D based
on a power curve fit was r = 0.96 for normal subjects
and r = 0.83 for diabetic patients when all vessels were
analyzed as independent data points (Fig. 2). When
the relationship between Q and D was investigated in
each subject separately, a positive correlation with an
average r of 0.98 ± 0.01 (range, 0.97-1.00) was observed in normal subjects and an average r of 0.93
±0.06 (range, 0.77-1.00) in diabetic patients. The
average r between D and Q was significantly smaller
than normal in diabetic patients (P < 0.01, by Wilcoxoh rank-sum test). Q varied as 6.1 • 10~6D287 in
normal subjects and as 1.4 • 10"5D267 in diabetic patients.
Values of total measured Q (QM), obtained by adding the Q of the four to five vessels in which BLDV
determinations were done in each eye, are summarized in Table 2 for normal subjects and in Table 3 for
diabetic patients. Average QM in diabetic patients was
approximately 20% larger than normal (P = 0.02, by
two-tailed, unpaired student t-test; Tables 2,3). Average ST of all visible veins in diabetic patients was also
TOTAL RETINAL DLOOD FLOW IN DIADETE5 / Grunwald er al
No. 2
Table 2. Total venous cross-section (ST), total
measured volumetric blood flow (QM), and
corrected total volumetric blood flow (QT)
in normal subjects
Subject
# Veins
Q measured
1
2
3
4
4
5
4
5
6
4
7
4
8
5
5
5
5
5
9
10
11
12
4
4
Mean
SD
ST
(cm2 X W~5)
QM
Qr
%ST
(iil/min)
(id/min)
95
100
100
100
100
87
100
100
100
59.8
66.8
62.5
87.7
88.3
105.7
63.5
93.9
79.3
101.2
86.1
91.5
82.2
15.7
25.9
31.6
35.1
43.0
35.7
44.4
33.3
42.5
36.7
38.5
37.3
42.2
37.2
27.3
31.6
35.1
43.0
35.7
50.6
5.4
6.2
92
100
100
97.9
4.1
33.3
42.5
36.7
41.5
37.3
42.2
38.1
%S T : Sum of cross-section of all veins from which volumetric blood flow
measurements were obtained expressed as a percentage of the cross-section of
all visible veins (ST).
Q T : Corrected value of Q M based on an estimation of flow rates for vessels
in which volumetric flow rate was not measured. Q T = QM/%Sr X 100.
significantly larger than normal, by approximately
30% {P = 0.001; Tables 2, 3).
The BLDV measurements of Vmax, and therefore
also of Q, were not obtained in all visible veins in all
eyes studied. The percent ST values shown in Tables 2
and 3 represent the sum of the cross-section of all
359
veins from which volumetric blood flow measurements were obtained expressed as a percentage of the
cross-section of all visible veins present in each studied eye. To include those vessels in which Q was not
determined, a total volumetric blood flow rate (QT)
was calculated based on an estimation offlowrates for
the vessels in which volumetricflowrate was not measured using the formula: QT = QM/%ST X 100 (Tables
2, 3). Average QT was also significantly larger than
normal in diabetic patients (P = 0.009; Tables 2, 3).
Nine patients with the mildest form of retinopathy
(level 20) already had an average QT of 45.6 ± 6.6
/Lil/min and ST of 103.3 ± 14.1 cm 2 -10" 5 ; these were
significantly larger than normal (P < 0.05 and P
< 0.01, respectively).
Significant positive correlations were observed between ST and QT in normal subjects (r = 0.88, P
= 0.0001, with a linear-regression equation of QT
= 9.11 -4- 0.35 ST) and in diabetic patients (r = 0.49, P
= 0.04, with a linear-regression equation of QT
= 22.49 + 0.23 ST).
Average D, Vmax, and Q for the largest retinal vein
in normal subjects and diabetic patients are shown in
Table 4. No significant differences from normal in D,
Vmax> or Q were observed when the measurements
obtained in the largest vein in diabetic patients and
normal subjects were compared.
We measured R in one of the major temporal veins;
it was significantly reduced in diabetic patients (P
Table 3. Total venous cross-section (ST), total measured volumetric blood flow (QM), and corrected total
volumetric blood flow (QT) in diabetic patients
Subject
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Retinopathy
level
#Veins
Q measured
%ST
40
45
50
20
40
20
45
20
20
20
30
20
40
20
20
40
40
20
4
4
4
4
4
4
5
4
4
5
5
5
4
4
4
4
5
5
100
82
92
89
96
100
100
93
94
94
93
94
100
88
100
94
100
100
Mean
94.9
SD
5.1
NS
Significance*
%Sr: Sum of cross-section of all veins from which volumetric blood flow
measurements were obtained expressed as a percentage of the cross-section of
all visible veins (ST).
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
ST
(cm2 X 10~5)
82.1
99.3
157.3
105.6
136.6
98.9
135.4
. 106.6
77.9
110.6
99.6
120.3
83.7
84.6
107.4
119.6
103.7
117.7
108.2
20.8
P = 0.001
QM
Qr
(til/mm)
(nl/min)
41.2
28.0
41.9
47.3
69.3
51.5
46.2
39.4
35.2
35.4
53.3
47.4
32.8
37.4
41.8
52.2
50.5
54.0
44.7
9.8
P = 0.02
41.2
34.1
45.6
53.0
72.4
51.5
46.2
42.3
37.6
37.8
57.4
50.3
32.8
42.3
41.8
55.3
50.5
54.0
47.0
9.7
P = 0.009
Q T : Corrected value of Q M based on an estimation of flow rates for vessels
in which Q was not measured. Q T = Q M /%S T X 100.
* Significantly different from normal by nonpaired student t-test.
360
INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / February 1992
Table 4. Average venous diameter (D), maximum
erythrocyte velocity (Vmax), and retinal volumetric
blood flow (Q) in the largest retinal vein of diabetic
patients and normal subjects
D(Mm)
Vmax (cm/sec)
Q Oil/min)
Normal
(n = 12)
Diabetes
(n = 18)
166 ± 13f
1.77 ±0.22
14.5 ± 2.6
179 ±20
1.72 ±0.37
16.2 ±3.9
Significance*
N.S.
N.S.
N.S.
NS = not statistically significant.
* Nonpaired student t-test.
tSD.
= 0.019, by unpaired student's t-test). Whereas Q decreased by 55 ± 9% during hyperoxia in normal subjects, it decreased by only 46 ± 11% in diabetic patients. Patients with more advanced retinopathy
showed smaller regulatory responses (Fig. 3), as demonstrated by the significant correlation present between R and retinopathy level (Spearman's rank
correlation coefficient = 0.52, P < 0.05). Although
average R for patients with mild retinopathy (level 20)
was smaller than normal (49 ± 11%), the difference
was not statistically significant. No significant correlation was found between QT and retinopathy level
(Spearman's rank correlation coefficient = 0.01,
In 16 diabetic patients, measurements of Q and R
in one of the major temporal veins were repeated a
few weeks later. The average percentage change in Q
between the two measurements was 2.8 ± 12%. The
difference in R between the two measurements
(R2(%) - R,(%)) was 1.9 ± 7.7%.
The distribution of Q in different areas of the retina
is shown in Table 5. On average, Q was significantly
larger in the temporal retina than in the nasal retina
Vol. 33
Table 5. Average of the sum of the venous
volumetric flow rate, Q Oul/min), in the temporal,
nasal, superior, and inferior regions of the retina
Normal
(n = 12)
Diabetes
(n = 18)
Temporal retina
Nasal retina
Significancef
25.0 ± 4.8
12.8 ±6.2
P = 0.0008
28.9 ± 8.5
15.8 ±7.0
P = 0.0002
NS
NS
Superior hemiretina
Inferior hemiretina
Significance!
19.0 ±3.6
18.4 ±3.7
23.7 ± 4.4
21.0 ±6.3
P = 0.03
P = 0.005
NS
NS
Significance*
NS = not statistically significant.
* Nonpaired t-test comparing flow in normal subjects and diabetic patients.
f Paired t-test comparing flow in the temporal and nasal or the superior
and inferior retina.
by 95% in normal eyes (P = 0.0008) and by 83% in the
diabetic eyes (P = 0.0002). In addition, Q in the superior retina was significantly larger than that observed
in the inferior retina in diabetic eyes (P = 0.03). In
normal subjects, however, this difference was not statistically significant (P = 0.48). When normal subjects
and diabetic patients were grouped together, Q was
also significantly larger in the superior retina than in
the inferior retina (P = 0.02).
We attempted to estimate total volumetric blood
flow rate from blood flow measurements obtained in
only one major retinal vein and from the total venous
cross-section using the following two procedures: (1)
Q was measured in the largest vein and scaled by a
factor consisting of the total venous cross-section/
largest vein cross-section (the result was defined as
Qnaisest) or (2) Q was measured in the second largest
vein and scaled by a factor consisting of the total venous cross-section/second largest vein cross-section
(QTsecond largest)- B o t h Q T l a r g e s l a n d QTsecond largest Were On
average about 15% larger than QT both in normal and
diabetic eyes (Table 6). In addition, a highly significant correlation was found between QT and Qriargest (r
60-
Table 6. Average corrected total volumetric blood
flow (QT) and estimated total volumetric blood flow
rates from the largest (QriargestX a n d the second
largest (QTsecond largest) retinal vein
50-
40 ••
3020 J20
30
40
Retinopathy Level
45
50
Fig. 3. Relationship between the retinal vascular regulatory response to 100% oxygen breathing (R) and retinopathy level of the
study eye. Spearman's Rank Correlation, rs = 0.52, P < 0.05. Average R in normal subjects was 55% ± 9%.
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
Normal eyes
Diabetic eyes
Qr
(nl/min)
\cTlargest
\J-Tsecond largest
(fil/min)
(nl/min)
38.1 ±6.2*
47.0 ± 9.7
43.1 ±9.4
54.6 ± 13.0
43.4 ± 8.9
54.4 ± 13.6
QTUIJCSV Estimated total volumetric blood flow from Q measured in the
largest vein scaled by a factor Total Venous Cross-Section/Largest Vein
Cross-Section.
QTsecond iarsrai: Estimated total volumetric blood flow from Q measured in
the second largest vein scaled by a factor Total Venous Cross-Section/Second
Largest Vein Cross-Section.
*SD.
No. 2
TOTAL RETINAL BLOOD FLOW IN DIABETES / Grunwold er ol
= 0.89, P = 0.0001, with a linear-regression equation
of Q T = 10.65 + 0.656-Q Tlargest ) or QTsecond u^est (X
= 0.87, P = 0.0001, with a linear-regression equation
of Q T = 11.59 + 0.637 • Q Tsecond lareest ).
Discussion
We attempted to measure total retinal venous
blood flow in patients with poorly controlled diabetes
mellitus and background diabetic retinopathy. Our
results suggest that, in these patients, total retinal volumetric blood flow rate is significantly larger than normal by approximately 23% (Table 6).
Measurements obtained in the largest retinal vein
in each eye showed an average Q that was only 12%
higher than normal, and this difference was not statistically significant (Table 4). This was similar to the
results in our previous study,8 in which average Q obtained in a single retinal vessel of patients with background retinopathy was approximately 8% above normal (also not statistically significant).
The results of our current study also show largerthan-normal veins in diabetic patients (Table 3), a
finding that agreed with previous reports demonstrating vasodilatation in diabetic patients.818 Both QT
and ST were already significantly larger than normal
in the nine patients with the mildest form of retinopathy (level 20).
However, Vmax was not significantly different from
normal in poorly controlled diabetic patients with
background retinopathy (Table 4). These results differed from those in our previous study8 in which a
decrease in average Vmax of 18% was observed in eyes
with background diabetic retinopathy.
There were important differences between the two
studies that might explain the discrepancy in Vmax results. In our current investigation, most of the patients had mild retinopathic changes (Fig. 3); in our
previous study,8 approximately one half of the subjects had severe background retinopathy. Because
Vmax seems to decrease with disease progression,8 a
more prevalent presence of severe background retinopathy could be related to the lower than normal
Vmax value found in our previous study.
In our current study, we selected only type I diabetic patients with poor glycemic control. Most of the
patients had GHb levels that were three standard deviations or more higher than the mean for nondiabetic subjects. In our previous study, GHb levels were
not measured, and no attempt was made to select patients specifically with poor control. It is possible that
patients with poor glycemic control could have a
higher Vmax than those with better control. This hypothesis was supported by our previous article,9 showing that in very poorly controlled diabetic patients
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
361
during marked hyperglycemia, Vmax is significantly
above normal and that normalization of blood glucose with insulin administration results in significant
decreases in Q and Vmax.
We stress, however, that the actual blood glucose
levels at the time offlowmeasurements in our current
study were, on average, close to those observed in our
previous one.8 Further studies are needed to clarify
the importance of the influences on the retinal circulation of chronically elevated blood glucose levels versus acutely elevated blood glucose levels.
As shown by our results, total retinal volumetric
blood flow was significantly larger than normal in patients with poorly controlled diabetes mellitus and
background retinopathy. This increase was caused
mainly by vasodilatation, which probably constitutes
a regulatory response of the retinal vasculature attempting to normalize a reduced supply of nutrients
or an increased concentration of metabolic waste
products in the retinal tissue.
The average r between Q and D in diabetic patients
was significantly smaller than that observed in normal
subjects. Moreover, among diabetic patients, there
was also a greater heterogeneity in r values, as demonstrated by a larger standard deviation. The reason for
this effect was not clear, but the changes in retinal
architecture and/or blood rheology, known to occur
in patients with diabetic retinopathy, might alter the
normal relationship between Q and D, leading to a
weaker or perhaps different association between these
parameters. We found no correlation, however, between the r values and retinopathy level or disease
duration, and therefore, we could not conclude that
more advanced diabetic pathology is associated with a
weaker correlation between Q and D.
In our previous study, in which measurements were
obtained in a single major retinal vein of each eye, a
large variability in Q was found in diabetic (12.3 =h 5.5
iul/min) and normal (11.3 ± 3.4 ^1/min) eyes.8 This
variability mainly was related to the differences in D
of the veins measured and the retinal areas perfused
by these vessels. Because of this variability, the sensitivity of the technique to detect differences in Q between groups of patients was approximately 23%, ie,
only changes larger than 23% could be demonstrated
in that study.
In our current study, we estimated QT by adding the
flows measured in the four to five largest retinal veins
and estimating theflowsin the few vessels in which Q
could not be measured. The variability in QT between
subjects was smaller than that observed for the largest
retinal veins (Table 4) and than seen in our previous
study.8 This intersubject variability in QT may be explained by the variability in ST because both quantities correlate significantly in diabetic patients (P
362
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1992
= 0.04) and normal subjects (P = 0.0001). Although
total retinal flow was significantly larger than normal
in the diabetic eyes, the difference in flow in the larger
retinal vein was not statistically significant.
From determinations of Q obtained a few weeks
apart in each of 16 diabetic subjects, we calculated
that our technique had a sensitivity of approximately
10%. Thus, we could detect an average change in Q of
10% or larger {P < 0.01, by two-tailed paired student
t-test). Two determinations of R done a few weeks
apart in 16 diabetic patients showed an average difference in R (R2(%) - R,(%)) of 1.9 ± 7.7%. From these
values, we determined that our technique had a sensitivity to detect average changes in R of approximately
6% (P < 0.01, by two-tailed paired student t-test),
when applied as described in a group of 16 diabetic
patients. These values are important for the design of
experimental protocols to study the effects of disease
or treatments on retinal hemodynamics.
We found R was significantly reduced in diabetic
patients, as reported previously.6 In addition, the significant decrease in R with progression of background
diabetic retinopathy (Fig. 3) further strengthened our
previous hypothesis that the decreased vascular regulatory response to hyperoxia may be related to the degree of retinal hypoxia.6-9"11
Measurements of QT are laborious and time consuming because determinations of Vmax must be made
separately in each individual major retinal vein. In
large-scale clinical studies, it may not be possible to do
all these measurements. We attempted to estimate QT
from the measurement of Q obtained in the largest or
second largest retinal vein and S-r (Table 6). This estimation was close to the actual value (higher by 15%,
on average), and furthermore, there was a strong
correlation between these quantities, suggesting that
this method may provide a rough estimation of total
retinal flow when actual total flow cannot be measured.
Average total retinal volumetric bloodflowsin normal subjects were close to the average of 34 ± 6.3
/zl/min reported previously15 in seven normal eyes.
Retinal volumetric bloodflowrates in individual major retinal veins were also similar to those reported
previously by our laboratory.8 These values of total
volumetricflow,however, are about 50% smaller than
those reported by others19 using laser Doppler velocimetry. The exact source of this discrepancy has not
been identified and probably stems from differences
in the way the technique was used.
The significantly larger Q observed in the temporal
retina than in the nasal retina of normal subjects
agreed with previous results.1519 A similar difference
also was found in our study in diabetic patients (Table
5). These differences can be explained by the larger
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
Vol. 33
area of the temporal retina. In addition, increased retinal thickness and perhaps an increased metabolic rate
in the macular region also may explain this finding.
By comparison with the inferior retina, the superior
retina showed a significantly larger Q when normal
subjects and diabetic patients were grouped together.
The inferior visual field is larger than the superior
visualfield.20In addition, the fovea is located slightly
below the center of the optic nerve head.21 These two
facts suggest that there may be a physiologic explanation for the differences in flow observed between the
superior and inferior retina.
Our results suggest that Q varies with D at a power
of 2.87 in normal eyes and 2.67 in eyes of diabetic
patients (Figs. 1, 2). The difference between these two
values was not statistically significant, and therefore,
we cannot conclude at this time that diabetes mellitus
affects this factor. A vascular tissue in which Q varies
with D at a power of 3 is the ideal system in terms of
the amount of energy needed to push blood through
the vessels.22 Any change from 3 represents a departure from the ideal situation.
Key words: retinal volumetric bloodflow,laser Doppler velocimetry, diabetic retinopathy, vascular regulation, erythrocyte velocity, hyperoxia
Acknowledgments
The authors thank Sharon Grunwald and Jim Gowan for
statistical consultation and analysis of the data and Dolly A.
Scott for preparation of this manuscript.
References
1. Hickam JB and Sieker HO: Retinal vascular reactivity in patients with diabetes mellitus and with atherosclerosis. Circulation 22:243, 1960.
2. Kohner EM, Hamilton AM, Saunders SJ, Sutcliffe BA, and
Bulpitt CJ: The retinal blood flow in diabetes. Diabetologia
11:27,1975.
3. Blair NP, Feke GT, Morales-Stoppello J, Riva CE, Goger DG,
Collas G, and McMeel JW: Prolongation of retinal mean circulation time in diabetes. Arch Ophthalmol 100:764, 1982.
4. Sinclair SH, Grunwald JE, Riva CE, Braunstein SN, Nichols
CW, and Schwartz SS: Retinal vascular autoregulation in diabetes mellitus. Ophthalmology 89:748, 1982.
5. Yoshida A, Feke GT, Morales-Stoppello J, Collas GD, Goger
DG, and McMeel JW: Retinal blood flow alterations during
progression of diabetic retinopathy. Arch Ophthalmol
101:225, 1983.
6. Grunwald JE, Riva CE, Brucker AJ, Sinclair SH, and Petrig
BL: Altered retinal vascular response to 100% oxygen breathing in diabetes mellitus. Ophthalmology 91:1447, 1984.
7. Feke GT, Tagawa G, Yoshida A, Goger DG, Weiter J, Buzney
SM, and McMeel JW: Retinal circulatory changes related to
retinopathy progression in insulin-dependent diabetes mellitus. Ophthalmology 92:1517, 1985.
8. Grunwald JE, Riva CE, Sinclair SH, Brucker AJ, and Petrig
BL: Laser Doppler velocimetry study of retinal circulation in
diabetes mellitus. Arch Ophthalmol 104:991, 1986.
No. 2
TOTAL RETINAL DLOOD FLOW IN DIABETES / Grunwold er ol
9. Grunwald JE, Riva CE, Martin DB, Quint AR, and Epstein
PA: Effect of insulin-induced decrease in blood glucose on the
human diabetic retinal circulation. Ophthalmology 94:1614,
1987.
10. Grunwald JE, Brucker AJ, Petrig BL, and Riva CE: Retinal
blood flow regulation and the clinical response to panretinal
photocoagulation in proliferative diabetic retinopathy. Ophthalmology 96:1518, 1989.
11. Grunwald JE, Brucker AJ, Schwartz SS, Braunstein SN, Baker
L, Petrig BL, and Riva CE: Diabetic glycemic control and retinal blood flow. Diabetes 39:602, 1990.
12. Riva CE, Petrig BL, and Grunwald JE: Retinal blood flow. In
Laser Doppler Flowmetry, Shepherd AP and Oberg PA, editors. Boston, KJuwer Academic, 1990, pp. 370-374.
13. Chase HP, Jackson WE, Hoops SL, Cockerham RS, Archer
PG, and O'Brien D: Glucose control and the renal and retinal
complications of insulin-dependent diabetes. JAMA 261:1155,
1989.
14. Klein BE, Moss SE, and Klein R: Longitudinal increase of glycemic control and diabetic retinopathy. Diabetes Care 10:273,
1987.
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
363
15. Riva CE, Grunwald JE, Sinclair SH, and Petrig BL: Blood velocity and volumetricflowrate in human retinal vessels. Invest
Ophthalmol Vis Sci 26:1124, 1985.
16. Sliney DH and Freasier BL: Evaluation of optical radiation
hazards. Applied Optics 12:1, 1973.
17. Damon DN and Duling BR: A comparison between mean
blood velocities and center-line red cell velocities as measured
with a mechanical image streaking velocimeter. Microvasc Res
17:330, 1979.
18. Skovborg F, Nielsen A, Lauritzen E, and Hartkopp D: Diameter of retinal vessels in diabetic and normal subjects. Diabetes,
18:292, 1969.
19. Feke GT, Tagawa H, Deupree DM, Goger DG, Sebag J, and
Weiter J: Bloodflowin the normal human retina. Invest Ophthalmol Vis Sci 30:58, 1989.
20. Anderson DR: Perimetry With and Without Automation. St.
Louis, CV Mosby, 1987, pp. 6.
21. Bishop PO: Binocular vision. In Adler's Physiology of the Eye,
Moses RA, editor. St. Louis, CV Mosby, 1975, p. 561.
22. Mayrovitz HN and Roy J: Microvascular bloodflow:Evidence
indicating a cubic dependence on arteriolar diameter. Am J
Physiol245:H1013, 1983.